In a Gallup poll in May, 71% of US adults said they would oppose a new AI data center being built in their area. Only 53% said they would oppose a new nuclear power plant. People would rather live next to a reactor than a server farm.
And yet ChatGPT hit a billion monthly users that same month, according to the market analysis firm Sensor Tower.
Both of those things are true at once. Right now, much of the world feels that way about AI: it loves the tools and hates them in roughly equal measure.
A self-loathing AI company
Over the summer I spoke with the CEO of Springboards, a startup building an LLM designed to generate a wider variety of responses than its mainstream rivals. Early in the call he said something I still can’t shake.
“We often say that we’re a self-loathing AI company. We don’t know if we really like what we’re doing,” the CEO of Springboards said.
I joked back that this made me a self-loathing AI journalist. I like my job. What I don’t like is much of what the technology I cover has turned into. It’s bent by hype, steered by zealots and impossible to avoid.
Plenty of people feel the same, and you can see it in the polling.
The numbers keep getting worse
According to the Pew Research Center, more US adults expect AI to have a negative impact on them personally, and on society in general, than expect a positive one. Young people are the most pessimistic.
Outside the US the mood isn’t much better. A Stanford University report found that more than half of people worldwide say AI products and services make them nervous. And in an NBC poll in March, AI polled worse than ICE.
That’s a rough set of numbers for an industry that wants to rebuild the economy around its products.
Usage keeps climbing anyway
Google DeepMind's Gemini isn’t far behind ChatGPT, with 950 million users logged in July. Pew found that half of US adults now say they use a chatbot. That’s more than twice the share who said so in 2023, and one in four say they use one every day.
The habit is global too. More than a third of adults across all 38 OECD countries, a group of the world’s wealthiest democracies, report using generative AI tools in the last three months.
So a lot of people dislike AI, and a lot of people can’t get enough of it.
The same people, most likely
One explanation is that these are two separate groups. I doubt it. With usage this high, the math only works if the Venn diagram is closing into a circle.
Another possibility is that the more people use AI, the less they like it. There’s at least a correlation pointing that way. The Global North, where adoption is highest, leans pessimistic. The Global South, where adoption is lower, is more optimistic.
My own read is that people are mostly angry at the companies. The resentment is aimed at the relentless push to wedge AI into every possible corner of our lives. These same companies keep telling us to brace for the biggest social and economic upheaval in generations. Anxiety is a reasonable response to that message.
We’ve watched this movie before
Social media played out the same way over the last 20 years. Billions of people signed up for Facebook and Twitter while the techlash against the companies running them kept growing. Google search followed the same pattern.
With social media, users had little leverage. If you quit a platform, you lost your content and your connections. You could put up with it or start from zero.
AI might go differently. There’s clearly more political appetite for regulation this time. All 50 US states have either passed or proposed laws governing how AI is developed and deployed. That adds up to a patchwork of more than 2,100 bills nationwide, a tenfold increase in three years.
There’s also real competition. Several top-tier open-source alternatives to Google, OpenAI and Anthropic are already available, which means more consumer choice and market pressure, at least for now.
No walking it back, but room to steer
I’m not naïve about this. Trillion-dollar companies are hard to move. But what comes next isn’t as inevitable as those companies like to suggest.
When I asked the Springboards CEO why his company was building a new model if he didn’t much like AI, he told me there was no walking back from LLMs. You could still make them do something different, though.
I’d like to see more of that. I want AI that’s upfront about what it can and can’t do, and that stops acting like it’s about to take over the world. That might do more to fix the polling than another billion users would.
